RBS applied to 3D nanostructures

Abstract

Due to recent use of nm scale 3D-devices in the semiconductor industry, there is an urgent need for metrology for such structures. Rutherford backscattering spectrometry (RBS) is a quantitative, traceable, and highly accurate technique to probe the composition of samples at the surface and in-depth up to a few microns, commonly used to characterize planar devices. Our research aims at extracting the nanometer-resolved 3D compositional information using RBS, even though the probing beam does not reach the spatial dimensions of the devices. Instead of relying on a focused beam to generate a localized analysis in a 3D-device, this is achieved by analyzing a large ensemble of identical devices. The ensemble analysis allows to reach a drastically improved signal intensity and hence sensitivity. We exploit the energy loss effects of the ions within the nanostructures to retrieve site-specific information at the nanoscale. In this way, we determined the local areal density on the top, sides, and bottom surfaces of an array of fins with a width of 35 nm and nominal pitch of 90 nm with a precision of a few percent, and limit of detection of 1013 atoms/cm2. Remarkably, the approach also allows to extract in-plane geometrical characteristics of the nanostructures like width, pitch, or under-etching. As a first example, we probe the selective removal of the SiGe layers in Si75Ge25/Si multi-layered fins with a width of 100 nm and pitch of 420 nm. We show that the Ge-signal in the RBS spectra correlate closely with the dimensional changes due to the applied selective etch of the Si75Ge25. The second example is an array of SiO2 fins with a width of 35 nm and nominal pitch of 90 nm. The silicon signal in the RBS spectra shows distinct features which can be directly related to the pitch and the width of the structures. Complementary simulations of theoretical targets with varying pitch and width support this interpretation and demonstrate the sensitivity to these parameters.